[Olsr-dev] Migration from olsrd v1 to v2

Ferry Huberts (spam-protected)
Mon Jul 27 21:39:54 CEST 2015



On 27/07/15 20:49, Bastian Bittorf wrote:
> * Ferry Huberts <(spam-protected)> [27.07.2015 20:42]:
>> IMHO you shouldn't abuse olsr (a routing protocol) to distribute user data.
>
> i'am not sure about this "conservative" opinion.

How is this conservative?
What the heck are you talking about?

> IMHO only the proto itself is aware of the best way how to
> distribute the data. it's the same discussion about abstraction
> of network layers: on the old days there was a consensus that
> layerX should not need to know anything about layerY.
>
> now we know that layer3 need to know something
> about layer1+2 for better routing...

You misunderstand.
You explained it partly yourself: layer X can use services of lower layers.
What you left out is that layer X can NOT use services of higher layers.

user data is a higher layer than the routing protocol and should 
therefore be on top.


The purpose of a routing protocol is to fill the OS routing tables.
It is NOT olsr that know best how to get data from A to B, it is the OS.

Ofcourse there are nuances, we can talk days about this. Let's not do 
that here.


>
> this means: when we distribute data with standard tools
> (e.g. multicast) it will not work in a mesh network. what
> are the "products" you have in mind?
>

See my other email

-- 
Ferry Huberts




More information about the Olsr-dev mailing list